Windows is a pain. Installed WSL ( linux sub ) and followed the procedures here:
This is ridiculous and why donât they add this thing to the official repo?
32 / 4 threads / 128 buckets / 4000 ram
Phase 1 > 4955.795
Phase 2 > 3674.756
Phase 3 > 4186.455
Phase 4 > 475.735
Total > 13291.746 = 3:41
THIS IS MASSIVE, my best time was 5:04
And these plots check out with plot check -n 100 ?
Well looking to check on that but I just noticed that the name structure is different
Plots are the same. In this optimized chiapos he did only asynchronous I/O in phase 3. The math is not changed.
1 Like
So I installed the .whl file that OP on Reddit kindly gave us in my virtual environment and changed chiapos dependency in the setup.py file. I confirm that this works correctly with my plot manager. But Iâm struggling to understand how to get the combined fork with all of the optimizations working correctly. It looks like I have to compile it myself which then requires the use of custom commands for it to work. How would this work with a plot manager like Plotman?
1 Like
I believe this is the case as well. I believe youâd have to edit the Plotman files to call the custom commands instead of the standard chia command.
1 Like
You will see it in logs
https://chiaforum.com/t/has-anyone-tried-this-speed-improvements/5479/16?u=gladanimal
Total time = 17666.812 seconds. CPU (206.100%) Sat May 29 23:21:41 2021
me likey 32/12 threads/ 128 buckets/ 10000 ram
Speed improvements are great, but as you can see with every of them cpu usage goes up. In other words, if you plot single (or small number of parallel) it will be faster, but if you already using your cpu at 100% to plot many in parallel i doubt it will be more plots per day.
Holy mother ofâŚ
Time for phase 1 = 3547.117 seconds. CPU (272.800%)
Time for phase 2 = 2717.627 seconds. CPU (100.390%)
Time for phase 3 = 3396.304 seconds. CPU (145.070%)
Time for phase 4 = 334.129 seconds. CPU (101.780%)
Total time = 9995.179 seconds. CPU (176.800%)
chia.plotting.check_plots : INFO Proofs 967 / 1000, 0.967
This means itâs legit?
That was a single plot on a 5900x, T1 = 2x2TB nvme RAID 0, T2 = 1.6TB Intel S3500, 6 threads, 4000 mem. Ubuntu 20.04
Unreal. All I had to do was learn Linux over the course of two days. WORTH IT! 
3 Likes
Oh BTW my test above was on windows
Forgot to say that, imma add the improvements soon to my Linux plotters
Was it Windows WSL ?
No, Windows using swar, I self built the binary for windows with the improvements above and the fallocate/allocate patch with some changes for it to allocate above 4gb files
https://puu.sh/HKTia/0cf1da0b5d.pyd
Herr the binary, but donât just trust me, I may publish the code later on github if I got some time
I basically just installed the correct python version using scoop and added the python dir to cmake then built the library with clions cmake support
The plots would check fine but if these optimizations rely on the Hellman Attacks as described in the âChia Proof of Space Constructionâ paper then a trade off between either more CPU and in memory space for the harvester, or potentially failing some of the proofs (but certainly not all) would have to be made.
1 Like
Well itâs consistent for me but the plots seems to be wrong or I made a mistake along the way. I compiled and build from his Chiapas and created them using ProofOfspace and the plots naming comes out incorrect. Now that alone should not be a problem but when you check them they throw an error
if u used proof of space, u need to use the correct memo else they broken
only 2 kinds of memos are supported atm by the plot checker afaik, (pool plots and regular plots, they detected by the byte size of the memo afaik)
also the names of the plots are always different for proof of space exe, thats fine, the chia daemon sets them normally chia-blockchain/chia/plotting/create_plots.py at main ¡ Chia-Network/chia-blockchain ¡ GitHub
Are this plots salvable in any way? Is it possible to get them to recognize them as valid?
TIA
No, else we would be able to convert normal plots to pooling plots 
1 Like
Made some changes and im now getting âsameâ speed increase on with ubuntu under WSL almost 4h
1 Like